Saturday, June 4, 2011

04/06/2011: The 'Confidence' trick?

Updated below: In the update below I address one particular point raised by some readers of this blog relating to PMIs and my analysis of these.


As promised in the earlier post, for the fans of the 'If only we were confident in Ireland' school of economic thinking... The school of thought, also known as the 'Green Jerseys', maintains that if confidence is high, then growth and employment will follow, so to get Ireland out of the crisis, positive thinking is needed.

Let's take a look at the data.

Keep in mind that we only have data for the period of May 2000-present and only for Services sector. Of course, Services is the largest sector in the Irish economy and it is more labour intensive, so the conclusions drawn from these observations should be expected to remain broadly valid for Manufacturing as well.

If the 'Confidence' thesis holds, we should expect some strong relationship between Confidence reading in PMIs and employment sub-index of the very same PMIs as well as PMIs main index which captures activity. This relationship might be subject to lags, of course, as Confidence sub-index is self-assessment of the future some 12-mo in advance, while Employment sub-index reflects current staffing levels, and the core PMI reflects current activity. Now, keep in mind that the 12-mo in advance expectations is for a continuum, not spot, in other words, growing confidence means expectations for improving business over the 12 months horizon.

First, consider whether there is a coincident relationship between Confidence and PMIs and Employment sub-indices:
Yes, there is a strong positive relationship between expansion signaling readings of confidence today the future and current levels of economic activity as measured by employment and PMIs. In other words, things tend to be optimistic (pessimistic) when PMIs are booming (shrinking) and employment is rising (falling).

Sounds like the 'Confidence' theory working? Not really - what's happening here is that when things are great, we expect them to stay great, on average. Alternatively, when things are bad, we expect them to stay bad for some time ahead. Which, of course is consistent with the fact that data we have covers 2000-present - two periods of pretty much persistent boom and then bust.

So let's take a look at change from month to month.
  • Does change in confidence imply change in current PMIs and employment? (If the 'Confidence' theory is right - it should, as future expected changes in activity should have a positive growth effect on current activity)
  • Does change in confidence today imply a change in future PMIs and employment? (If the theory is correct, then it should, with some lag kick in in terms of positive real outcomes)

Sorry, but it appears that a change to higher Confidence in the future in any given month relative to previous month has virtually no relation to either present or future Employment changes or future PMIs. It has a tiny positive connection to present PMIs, however, but barely enough to be called 'significant' from statistical point of view. In other words, we might get all giddy chirpy about the great future we have, and yet it will be unlikely (highly unlikely) - according to the PMIs data - to translate into significant gains in either services activities or employment, neither today, nor in the near future (I tested longer lags up to 12mo and the results do not change by much).

This of course does not mean that positive sentiment is not a good thing for the economy. If positive sentiment is backed by something more tangible - reforms, improved exports, growth in consumer or investor confidence - some real productive fundamentals, then of course it will matter. But that is not the 'Confidence' theory. The 'Confidence' theory says 'negativity hurts economy'. No, folks - it doesn't. You can't talk yourself into a recession. And the 'Confidence' theory claims that if we get 'positive' about the future, things will improve (presumably improve significantly, otherwise, what's the point). This is not what the data is showing.

So what's going on, then? We know that Confidence is associated with performance, but we also now know that at least in Ireland, over the period looked at, changes in confidence are not associated with changes in performance either today or in the future.

Of course, Ireland is a small open economy. Which means it is volatile and is subject to constantly shifting external 'winds' of change. May it be the case that 'Confidence' theory doesn't work in Ireland because our real economy is subject to external forces and shocks? Ok, let's test this proposition. Let's control for contemporaneous backlog of orders, leaving only that component of Confidence that is not influenced by these backlogs. In other words, let's consider that part of our self-assessed optimism (pessimism) that is unrelated to the actual observed increases in new orders (decline in these orders). Furthermore, let's slightly smooth the series tor educe volatility by using a 2mo moving average on all variables.
An interesting result above. The link between confidence and contemporaneous PMI and Employment is now virtually gone (compare the results with Chart 1 above), which exactly supports my conclusions made following Chart 1. What matters to the turning of the economy, folks, is the real economic activity - rising backlogs, new orders, new export orders. What doesn't matter much at all, it appears, is 'Confidence'. So, please, go on, feel great - it might improve your smile, your utility, your view on life - all of which are great results. But don't hold much hope that it will improve the economy and reduce unemployment.

In the end, to achieve these two objectives, we need new businesses to be created, new markets to be accessed, new products and services to be developed and marketed, and so on, and new reforms implemented. Unfortunately, the 'Confidence' theory can lead us into complacency of avoiding making hard choice to have such reforms, to support our entrepreneurs, our companies and workers.

Ignoring the rain might make getting wet tolerable or even fun, but it won't make you any less soaked.

Update: Some websites contain references to these series of posts on PMIs. In particular, there is an occasional refrain to my view that (1) I would prefer seeing strong (above 60) readings in some sub-indices, and (2) my insistence that an 'improvement' in the sub-index reading from a number below 50 to another, higher number below 50 is not an improvement. Let me explain my views on these 2 points.
  1. Readings above 60 are rare, that is true. But PMIs refer to comparative/relative performance metrics. Now, real recovery is not, I repeat - not - associated with growth returning to a long-term trend, but growth overshooting long-term trend as economy goes from negative growth (contraction) to expansion. Thus, for example in the Services series, near-recession of 2001 is returned to growth by PMIs reaching for 60.8 by April 2002. In Manufacturing series contraction in October 2001 to 46.1 is returned to growth with June 2002 reading of 54.5 (so -3.1 from 50 to +4.5 - a ratio of 1.45), contraction of July 2003 (45.8) is returned to growth with a peak of 55.2 in June 2004 (-4.2 to +5.2 - a ratio of 1.24). Now, bottom of the latest contraction was at 32.6 which should be consistent - if we take the above two episodes averages (ratio of 1.35) - with a rise to above 67. We've gone up to 62 in March 2010, but we have not seen this translate into overall economic growth. Hence, my preference would be to see more episodes of 60+ readings in PMIs. Either way, all of the episodes we have on the record so far are episodes relating to either 'near recessions' or temporary declines in the series not associated with a recession at all. Except for the current crisis, that is. O course, this 60 is not a 'hard' target. Read carefully what I said (here): "Either way, of course, I'd rather see PMIs at above 60 reading, than heading for a downward territory". This is a statement of 'truism' - as in: I'd rather see things improve than get worse. Sadly, some anonymous commentators on some of the forums out there are not getting even this simple concept...
  2. When the series read below 50, the series show contraction. Thus, for example, a reading of 44 in one month followed by a reading of 46 in the next month does not mean that economy has improved from month to month. It means that the economy has deteriorated at a slower rate. If you are familiar with compounded effects of recessions (expansions), you would know that having a loss of 10% in month 1 followed by a 5% decline in the other month implies a cumulative decline of 14.5%. An improvement would be if following a 10% drop in on month, economy grows by, say, even 1% in the next month, thereby reducing the original decline to a cumulative decline of 9.1%. Let me quote Brad DeLong on this: "Getting worse more slowly is not the same as getting better".

2 comments:

Anonymous said...

Always surprised the papers report on this when there is real monthly activity data from CSO that hardly gets a mention not far behind it

Mark said...

@CG

Not to be picky but should your -3.1 number above not read -3.9 i.e. 46.1-50 = -3.9.